Partition function loop series for a general graphical model: free energy corrections and message-passing equations.
Jing-Qing XiaoHaijun ZhouPublished in: CoRR (2011)
Keyphrases
- belief propagation
- free energy
- message passing
- graphical models
- partition function
- approximate inference
- loopy belief propagation
- markov random field
- probabilistic inference
- factor graphs
- probabilistic model
- probabilistic graphical models
- random variables
- exact inference
- fixed point
- markov networks
- energy minimization
- conditional random fields
- graph cuts
- bayesian networks
- belief networks
- structure learning
- stereo matching
- distributed systems
- variational methods
- importance sampling
- structured prediction
- conditional independence
- image segmentation
- pairwise
- three dimensional